This is a very common picture. If there is something special, it may be generated by AI. There is nothing special about it. But after adding three positioning points to it , this picture becomes a QR code that can be scanned and recognized:
real or fake? If you don't believe me, long press the picture to identify it! The first time I saw it, I was amazed. QR codes can be so artistic? ! There are more examples below, let's take a look.
<<< Swipe left and right to see more>>>
Most of our common QR codes are like this. Compared with the above QR code, the aesthetics can be judged.
<<< Swipe left and right to see more>>>
<<< Swipe left and right to see more>>>
This is the picture QR code realized by the latest AI technology. The technical name is ControlNet for QR Code . It is an AI-generated QR code technology that can subvert the QR code industry.
So how did he generate it, let's take a look at the general principle and process. It is based on the open source product Stable Diffusion combined with a technology called ControlNet plug-in, through high-intensity AI training, it generates a QR code that looks like a picture. More technical details may be more difficult to understand, so I won't elaborate too much here.
Stable Diffusion is an open source AI text-to-image generation model. It can generate high-quality, high-resolution, and high-fidelity images only by inputting text prompts, as mentioned many times in previous articles.
ControlNet is a plugin for controlling AI image generation. It uses a technique called conditional generative adversarial networks to generate images. Unlike traditional GANs, ControlNet allows users fine-grained control over the generated images. This makes ControlNet useful in many application scenarios, such as computer vision, art design, virtual reality, and more. In short, ControlNet can help users precisely control the generation of AI images for better visual effects.
Now you must be thinking, how to generate your own QR code, is there an online website or App? Not yet, and the threshold for ordinary users is still relatively high. With the improvement of technology, I believe there will be more fool-like applications.
If there is a complete new product launch, I will recommend it to you, stay tuned.
Further reading: